Failure of nonstructural components during an earthquake can cause widespread property damage, lead to lengthy downtime of the structure, and pose a life-safety threat to the occupants. Current code provisions aim to minimize the life safety threat by specifying lateral force demands and anchoring requirements. These code requirements are based on a simplified equation that does not fully consider the component attachment’s contribution to its overall dynamic response. Research suggests that the component attachment is an important parameter that determines its dynamic properties. Most analyses of the dynamic response of floor-anchored nonstructural components use an idealized single-degree-of-freedom system with a concentrated mass. Nonlinear behavior is included in the component force-displacement relations while the boundary conditions remain fixed based to the floor. Assuming a nonlinear behavior of the attachment changes the fixed-based boundary conditions and results in uplift at the base of the model. Models that include uplift at the base can be found for analysis of rocking structures and soil-structure interaction, but are rarely applied for analyzing nonstructural components. This dissertation focuses on acceleration-sensitive equipment and examines the attachment design of floor-anchored mechanical and electrical nonstructural components. The dissertation includes three parts. (i) An extensive experimental shaking-table test study of a nonstructural experimental model that is attached via several attachment designs to a concrete slab. The experimental results demonstrate that the selected attachment properties govern the boundary conditions of the nonstructural component and strongly influence its dynamic response. (ii) A mechanics-based numerical modeling approach for floor-anchored nonstructural components attached via steel channel connections. It defines a generalized analytical approach to estimate the force-displacement relations of the attachment and offers a simple approach for the engineer to estimate the contribution of the attachment design to the dynamic amplification of the component. (iii) A numerical study that examines the mechanics-based modeling approach with a wide range of input motions, component parameters, and attachment properties.

Earthquakes that have occurred recently across the globe in various countries including United States, Japan, New Zealand, Haiti, Turkey and Italy have brought into light the poor seismic performance of older-type, non-ductile concrete buildings. These buildings, mainly designed and constructed prior to 1980s, lack proper seismic detailing and may pose an unacceptably high seismic risk.

Non-ductile concrete buildings pose one of the greatest seismic safety problems in the world due to the large amount of old buildings constructed in earthquake prone regions. It is indicative that according to the Concrete Coalition and the California inventory project there are 16,000-17,000 of these buildings in high earthquake risk counties of California. Many of these buildings have high occupancies, including residential, commercial and critical services. In case of a severe earthquake, the severe damage or even collapse that could occur in these buildings could result in large number of casualties.

While engineers generally recognize that proactive steps are required to address the risk posed by these buildings, mitigation efforts are largely stymied by insufficient knowledge about the scale of the problem, insufficient tools to identify the truly dangerous buildings, high costs of strengthening, and owner resistance to pay for the strengthening with uncertain benefits. This study constitutes an effort to identify seismically hazardous concrete frame buildings through simplified methods that do not require complicated analysis.

Three idealized concrete frame buildings with different heights are used as archetypes. The study attempts to link the collapse performance of these buildings with various structural deficiencies that appear commonly in older construction practice. To evaluate the performance of these buildings non-linear dynamic analysis for several far-fault ground motions is performed. The analysis considers nonlinearities associated with flexural yielding, shear and axial failure. The main deficiencies explored are development of weak story mechanisms due to strong column-weak beam designs, brittle shear or axial failure modes associated with inadequate column shear reinforcement detailing, and splicing and connectivity weaknesses between structural members.

The results indicate that the suggested methods can be used to assess the collapse risk of older-type concrete buildings. The methods developed in the current study use simple engineering parameters such as column-to-beam strength ratio and column flexural to shear strength ratio to estimate the collapse risk of older type concrete buildings. A probabilistic approach is suggested that takes into account record-to-record variability and could accommodate as well uncertainty associated with structural properties and collapse modeling.

In Chapter 7 the proposed methodology is evaluated by applying it to the three idealized buildings developed. The estimated probabilities of collapse calculated for each of the buildings according to the proposed methodology are compared with the values provided by sophisticated non-linear dynamic analyses. The results suggest that the proposed methodology successfully identifies deficiencies that are leading to high collapse potential and provides an effective tool in classifying collapse prone concrete frame buildings.

The intent of the strong-column weak-beam (SCWB) strength ratio in building codes is

to reduce the likelihood of the formation of story mechanisms in reinforced concrete special

moment-resisting frames subjected to seismic loading. Previous research has shown that

for tall buildings the current code requirement is not sucient to prevent these undesirable

plastic mechanisms from forming and leading to collapse of the structure. Furthermore,

nonlinear analyses of structures greater than four stories with strength ratios of 2:0 or greater

have shown story mechanisms to still occur. It is unclear whether complete prevention of

story mechanisms is possible or even necessary in tall buildings. To achieve a complete

building mechanism, the required SCWB ratio would lead to dimensions that would be

deemed unacceptable to project sponsors.

To determine the eects that SCWB strength ratios have on collapse mechanisms and

collapse capacities of buildings, several structures with diering SCWB ratios and heights

were analyzed: namely, 12-, 18- and 24-story structures each with 1:2, 1:4, 1:6, 1:8, and 2:0

SCWB ratios, for a total of 15 structures. Numerical modelling of these perimeter frame

structures was done in Opensees. A nonlinear static analysis and an incremental dynamic

analysis (IDA) using 30 ground motions were performed on each structure. Fragility curves

were obtained using the maximum likelihood method from the results of the IDA. The

probability of collapse given a maximum credible event, P(CjMCE), of each structure was

subsequently obtained. Maximum beam and column end rotations occurring during the IDA

were plotted to examine the types of mechanisms formed.

A laboratory test and analytical research program was undertaken to characterize the performance of reinforced concrete beams with high-strength reinforcement subjected to reversed cyclic lateral loading simulating earthquake effects. The beams are representative of beams used in special moment frames. Four beams were tested in the laboratory test investigation, one with A706 Grade 60 reinforcement, one with Grade 100 reinforcement having tensile-to-yield strength ratio (T/Y) of 1.17, one with Grade 100 reinforcement with T/Y = 1.26, and one with A1035 Grade 100 reinforcement. In each beam, the noted reinforcement grade was used for both longitudinal and transverse reinforcement, except for beam with Grade 100 T/Y = 1.17 that had transverse reinforcement of Grade 100 with T/Y = 1.26. Overall, all beams achieved rotation capacity of at least 0.045 radians. The beams with A706 Grade 60 and Grade 100 (T/Y = 1.26) reinforcement failed by buckling of longitudinal bars over several hoop spacings. The other two beams with Grade 100 reinforcement failed by fracture of longitudinal bars at the maximum moment section. Strain gauges installed on longitudinal bars indicated that beams with higher T/Y achieved greater spread of plasticity compared to beams with lower T/Y.

In the analytical study, the seismic performance of tall reinforced concrete special moment resisting frames with high-strength reinforcement was investigated through nonlinear dynamic analyses. Four 20-story reinforced concrete moment frames, three reinforced with Grade 100 steel and one with Grade 60 steel were designed in accordance with ASCE 7-16 and ACI 318-14 at a hypothetical site in San Francisco, California. All four frames had the same dimensions and concrete properties, resulting in identical design drifts. Frames with Grade 100 reinforcement were designed to have reduced amount of longitudinal reinforcement to provide equivalent nominal strength as was provided in the Grade 60 reinforcement model. Tests had demonstrated that frames with higher-grade reinforcement had greater strain penetration into beam-column joints, resulting in greater slip of reinforcement from connections. This effect combined with reduced reinforcement ratios caused the frames with Grade 100 reinforcement to be more flexible than the frame with Grade 60 reinforcement. In addition, some currently available types of Grade 100 reinforcement have lower tensile-to-yield strength ratio and lower uniform elongation compared with Grade 60 reinforcement. The reduced T/Y results in reduced strain-hardening, increased strain localization, and increased P-Delta effects. The effects of these local behaviors on overall frame performance are studied through the nonlinear dynamic analyses. The various types of reinforcement were found to result in minor differences in overall frame seismic performance.

Steel and precast columns are commonly designed to transfer moment loads to concrete foundations through cast-in-place headed anchors. In design office practice in the United States, connection strength has been evaluated considering mechanisms emphasizing joint shear, strut-and-tie modeling, or anchoring-to-concrete. For any given connection, the strengths calculated with these three methods can differ by a wide margin. The breakout failure mode is not routinely checked, even though recent physical tests have shown it can limit connection strength. Strategically placed shear reinforcing bars have been shown to increase the strength and displacement capacity of anchored connections governed by breakout failure. However, designers cannot consider the beneficial influence of shear reinforcement on breakout failure as it is generally ignored by current building codes (for example, ACI 318-19 and Eurocode EN 1992-4). Through physical testing and finite element simulations, this dissertation investigates the application of various design methods, including possible enhancements that improve strength estimates. A novel approach to calculating the concrete breakout strength with distributed shear reinforcement is proposed. This dissertation is divided into three sections: (i) a description of observations from physical tests that consisted of full-scale interior steel-column to concrete-foundation connections with details typical of current construction practice on the West Coast of the United States, (ii) a description of how physical test data are used to calibrate finite element models of column-foundation connections to investigate critical variables, and (iii) a discussion of a novel approach to calculating the concrete breakout strength considering the additive effect of concrete and reinforcing bars when distributed shear reinforcement is present.

Damage observed near the base of shear walls of reinforced concrete buildings after the Chile (2010) and New Zealand (2011) earthquakes are signs of shortcomings in the design of walls that need to be addressed. This investigation presents results of an experimental test program on ten reinforced concrete rectangular prisms representative of the flexural compression zone of flanged shear walls. The tested elements have transverse reinforcement detailing that matches or exceeds modern code requirements for special boundary elements. The main test variables were the amount and spacing (both vertical and horizontal) of the hoop and crosstie reinforcement. The elements were subjected to monotonically increasing axial compression until failure. Effects of strain gradient (both through the wall length and along the wall height) and effects of wall shear are not represented in the present tests. Nonetheless, the axial compression tests provide insights into the behavioral characteristics of actual wall boundaries. The global force shortening behavior of the specimens was commanded by a thin core which integrity was heavily compromised due to cover spalling, rebar buckling and out-of-plane instability. Measured load-displacement relations did not exhibit an acceptable ductile behavior suggesting that current building code requirements for special boundary elements do not necessarily achieve effective confinement to be protected against brittle axial failure. Enhanced detailing (increasing the volumetric ratio of confinement reinforcement and decreasing its horizontal spacing) improved behavior but did not produce ductile response in all cases. Reported damage extension concentrated over length corresponding to two-and-half times the thickness of the specimens. Compressive strain limits for stable behavior are proposed to be function of the gage length over which they are measured.

Bar buckling reduced the load carrying capacity of the reinforced concrete prisms because of the strength loss suffered by the longitudinal reinforcement, but also because it prevented the effective confinement of the concrete core. An experimental campaign comprising 48 analytical specimens allowed studying the relationship between tie spacing and stiffness, and the diameter of the longitudinal bars, that influenced their response when undergoing lateral instability (inelastic buckling). The behavior of tied bars undergoing lateral instability in the inelastic range is highly influenced by the relative restrictive tie spacing over which bar buckling is forced into, and the relative stiffness of the transverse ties and the longitudinal bar. The experiments assume a rigid contact between the bar and the tie, therefore hook opening is not modeled. For the range of tie stiffness and bar geometries tested, the results indicate that the tie spacing has to be smaller than 4.5 times the bar diameter to prevent bar buckling over a large range of plastic axial strains.

Empirical core stress strain curves, accounting for bar buckling, are reported for point wise strain measurements, as well as for average axial strains recorded within the damaged region. The results show that usable strain limits, to guarantee a stable core response in pure compression, are between 1.1 and 2.0%. Average empirical core stress strain curves are proposed for modeling purposes. Implication of the compressive strain limits observed are evaluated in a hazard-consistent manner by means of the Conditional Scenario Spectra (CSS). The CSS is a set of realistic earthquake spectra with assigned rates of occurrence that reproduce the hazard at a site. Structural responses are obtained by means of numerical analysis of a multistory shear wall under the seismic demand of more than eight-hundred ground motions consistent with the CSS. The case study allows estimating risk curves to evaluate the likelihood of exceeding certain threshold compressive strains in the boundary of the cross section. The single case numerical model showed that the limited strain capacity of these elements is only likely to negatively impact the behavior of the wall system at risk levels beyond the code-based expectations of good behavior.

An experimental and analytical research program was conducted to characterize performance of concrete beams longitudinally reinforced with high-strength steel subjected to monotonic loading. Performance was evaluated at both service and limit states.

Experimental tests were conducted in two series of four beams each. In each series, two beams were reinforced with Grade 60 steel and two with Grade 100 steel. All beams were designed to ultimately fail in flexure. In Series 1, failure of beams was expected at high longitudinal strains. For this series, the tensile-to-yield strength ratio of the longitudinal steel (T/Y) was varied to study its effects on beam performance. In Series 2, failure was expected at low longitudinal steel strains. The strain in the steel at the crushing strain of concrete was the variable under investigation for this test series. Load-deflection relations, strain distributions, and plastic deformations for beams in each series were compared with one another to evaluate relative performance. For the Series 1 tests with low reinforcement ratio and, hence, large reinforcement strain at failure, a reduction in T/Y caused a reduction in the spread of plasticity and, consequently, a reduction in displacement capacity. For the Series 2 tests with high reinforcement ratio, deformation capacity was reduced as net tensile strain was reduced. For a given value of net tensile strain, and for a given value of the difference between net tensile strain and yield strain, the beams with Grade 100 reinforcement had equal or greater deformation capacity than the beams with Grade 60 reinforcement. Considering service-level response, typical techniques for evaluating beam performance, such as crack widths and deflections in the elastic range, were adequate for assessing behavior of beams independent of steel grade.

Moment redistribution is a concept that allows designers to shift the static moment envelope by allowing for nonlinearity at certain sections. In the present study, an analytical program examined moment distribution capacity through finite-element models. Beams in the analytical study had longitudinal reinforcement of Grade 60 and 100, with T/Y and aspect ratios spanning typical design values, which were variables identified as critical for determining moment-redistribution capability based on a theoretical derivation. The boundary condition for all beams was fixed-fixed and transverse reinforcement was provided to allow the formation of a mechanism starting with hinging at the fixed ends followed by hinging at the beam center. Input parameters for the models evaluated as part of this study were informed by the results of the experimental investigation. The theoretical moment-redistribution calculation was demonstrated to be conservative based on the output from these analyses. A relationship bounding percent allowable moment redistribution based on net tensile strain is proposed for grades of steel higher than Grade 60.

ABSTRACT

Analytical and Experimental Assessment of Seismic Vulnerability of Beam-Column Joints without Transverse Reinforcement in Concrete Buildings

by

Wael Mohamed Hassan

Doctor of Philosophy in Engineering - Civil and Environmental Engineering

University of California, Berkeley

Professor Jack P. Moehle, Chair

Beam-column joints in concrete buildings are key components to ensure structural integrity of

building performance under seismic loading. Earthquake reconnaissance has reported the

substantial damage that can result from inadequate beam-column joints. In some cases, failure of

older-type corner joints appears to have led to building collapse.

Since the 1960s, many advances have been made to improve seismic performance of

building components, including beam-column joints. New design and detailing approaches are

expected to produce new construction that will perform satisfactorily during strong earthquake

shaking. Much less attention has been focused on beam-column joints of older construction that

may be seismically vulnerable. Concrete buildings constructed prior to developing details for

ductility in the 1970s normally lack joint transverse reinforcement. The available literature

concerning the performance of such joints is relatively limited, but concerns about performance

exist.

The current study aimed to improve understanding and assessment of seismic performance

of unconfined exterior and corner beam-column joints in existing buildings. An extensive

literature survey was performed, leading to development of a database of about a hundred tests.

Study of the data enabled identification of the most important parameters and the effect of each

parameter on the seismic performance.

The available analytical models and guidelines for strength and deformability assessment of

unconfined joints were surveyed and evaluated. In particular, The ASCE 41 existing building

document proved to be substantially conservative in joint shear strength estimation. Upon

identifying deficiencies in these models, two new joint shear strength models, a bond capacity

model, and two axial capacity models designed and tailored specifically for unconfined beamcolumn

joints were developed. The proposed models strongly correlated with previous test

results.

In the laboratory testing phase of the current study, four full-scale corner beam-column joint

subassemblies, with slab included, were designed, built, instrumented, tested, and analyzed. The

specimens were tested under unidirectional and bidirectional displacement-controlled quasi-static

loading that incorporated varying axial loads that simulated overturning seismic moment effects.

The axial loads varied between tension and high compression loads reaching about 50% of the

column axial capacity. The test parameters were axial load level, loading history, joint aspect

ratio, and beam reinforcement ratio. The test results proved that high axial load increases joint

shear strength and decreases the deformability of joints failing in pure shear failure mode without

beam yielding. On the contrary, high axial load did not affect the strength of joints failing in

shear after significant beam yielding; however, it substantially increased their displacement

ductility. Joint aspect ratio proved to be instrumental in deciding joint shear strength; that is the

deeper the joint the lower the shear strength. Bidirectional loading reduced the apparent strength

of the joint in the uniaxial principal axes. However, circular shear strength interaction is an

appropriate approximation to predict the biaxial strength. The developed shear strength models

predicted successfully the strength of test specimens.

Based on the literature database investigation, the shear and axial capacity models developed

and the test results of the current study, an analytical finite element component model based on a

proposed joint shear stress-rotation backbone constitutive curve was developed to represent the

behavior of unconfined beam-column joints in computer numerical simulations of concrete

frame buildings. The proposed finite element model included the effect of axial load, mode of

joint failure, joint aspect ratio and axial capacity of joint. The proposed backbone curve along

with the developed joint element exhibited high accuracy in simulating the test response of the

current test specimens as well as previous test joints.

Finally, a parametric study was conducted to assess the axial failure vulnerability of

unconfined beam-column joints based on the developed shear and axial capacity models. This

parametric study compared the axial failure potential of unconfined beam-column joint with that

of shear critical columns to provide a preliminary insight into the axial collapse vulnerability of

older-type buildings during intense ground shaking.

Structural (shear) walls are used worldwide to resist gravity and earthquake loads. In many countries, structural walls commonly are constructed with a rectangular cross section, or a cross section made up of interconnected rectangles, without an enlarged boundary element. In some countries, design practice has resulted in walls that are more slender than those used in the past. For example, in Chile and elsewhere it is not unusual to find rectangular wall edges having thickness of 6 to 8 in. (150 to 200 mm), resulting in floor-to-floor slenderness ratios reaching hu⁄b= 16 or greater. Such walls can be susceptible to overall wall buckling in which a portion of the wall buckles out of the plane. The main objective of this research is to develop a methodology for evaluation of the onset of lateral instability in reinforced concrete slender walls. First, a simplified buckling mechanics solution for prismatic columns under inelastic tension/compression cycles is presented and later evaluated using the results of column tests. Later, three numerical models for buckling in columns are evaluated: nonlinear beam-column elements with fibers, and two-dimensional and three-dimensional nonlinear finite element models. Wall boundaries have strain gradient along the wall length, which would tend to brace the edge of the wall with the result that the simplified mechanics solution may give an over-conservative estimation of the onset of buckling. The theory may also be conservative for walls where the axial force in the boundary elements is not constant along the unsupported height, as may occur where moment gradients occur. To study these effects, analytical models of columns and walls are implemented. A simple approach is proposed to reduce the over conservatism of the simplified mechanics in cases where strain gradients cannot be neglected. Later, two-dimensional nonlinear finite element models are used to analytically reproduce the experimental response of wall tests. These models are used to estimate strain profiles for evaluation of the onset of lateral buckling in slender boundaries. Strain profiles are also estimated from a plastic hinge model. Finally, three damaged buildings in Chile are analyzed using linear models for the buildings and nonlinear models for isolated walls. Both buildings had some buckled walls after the 2010 Maule earthquake. Based on these studies, it is concluded that buckling in Chilean buildings most likely was a secondary failure that occurred after initial crushing of the wall boundaries.